Data Efficient and Weakly Supervised Computational Pathology on Whole Slide Images

@article{Lu2020DataEA,
  title={Data Efficient and Weakly Supervised Computational Pathology on Whole Slide Images},
  author={Ming Y. Lu and Drew F. K. Williamson and Tiffany Y. Chen and Richard J. Chen and Matteo Barbieri and Faisal Mahmood},
  journal={Nature biomedical engineering},
  year={2020}
}
Deep-learning methods for computational pathology require either manual annotation of gigapixel whole-slide images (WSIs) or large datasets of WSIs with slide-level labels and typically suffer from poor domain adaptation and interpretability. Here we report an interpretable weakly supervised deep-learning method for data-efficient WSI processing and learning that only requires slide-level labels. The method, which we named clustering-constrained-attention multiple-instance learning (CLAM), uses… 

Detection of Prostate Cancer in Whole-Slide Images Through End-to-End Training With Image-Level Labels

It is shown that modern CNNs, trained using the streaming approach, can extract meaningful features from high-resolution images without additional heuristics, reaching similar performance as state-of-the-art patch-based and multiple-instance learning methods.

Deep Learning-based Computational Pathology Predicts Origins for Cancers of Unknown Primary

A deep-learning-based algorithm-Tumour Origin Assessment via Deep Learning (TOAD) that can provide a differential diagnosis for the origin of the primary tumour using routinely acquired histology slides and results in concordance for 61% of cases and a top-3 agreement of 82%.

Artificial Intelligence in Pathology: A Simple and Practical Guide

A simple and practical guide to AI methods used in pathology, such as digital image analysis, next-generation sequencing, and natural language processing is offered.

FEZ: Flexible and Efficient Zoom-In for Ultra-Large Image Classification

Experimental results show that FEZ is capable of delivering similar accuracy with higher speed than a recently proposed whole-slide training method (WST) and existing sampling methods such as Multiple Instance Learning (MIL) and CLAM, and is 42× and 22× faster than MIL on the Camelyon16 and TCGA Lung Cancer datasets.

Multimodal Guidance for Medical Image Classification

A lightweight guidance model is developed – an autoencoder-like deep neural network – that learns a mapping from the latent representation of the inferior modality to the latent representations of its superior counterpart to compensate for the absence of the superior modality during inference time.

Interpretable deep learning of myelin histopathology in age-related cognitive impairment

The results demonstrate a scalable platform with interpretable deep learning to identify unexpected aspects of pathology in cognitive impairment that can be translated to the study of other neurobiological disorders.

A Fully Automated Multi-Scale Pipeline for Oral Epithelial Dysplasia Grading and Outcome Prediction

This work finetuned HoVer-Net+ for the simultaneous segmentation of nuclei and the epithelial layers in heamatoxylin and eosin (H&E) stained whole slide images (WSIs) and employed a multi-scale attention-based multiple instance learning architecture for the prediction of OED status, grade, recurrence and malignant transformation.

Weakly-supervised learning for image-based classification of primary melanomas into genomic immune subgroups

This work develops deep learning models to classify gigapixel H&E stained pathology slides, which are well established in clinical workflows, into immune subgroups and shows that pathology-specific self-supervised models generate better representations compared to pathology-agnostic models for improving model performance.

A QuadTree Image Representation for Computational Pathology

This work presents a method to generate an interpretable image representation of computational pathology images using quadtrees and a pipeline to use these representations for highly accurate downstream classification, and shows it is highly accurate.

Attention-based Multiple Instance Learning with Mixed Supervision on the Camelyon16 Dataset

It is shown on the Camelyon16 dataset that even a small proportion of slides with pixel-wise annotations can improve their classification but also the localization of tumorous regions, which improves the consistency of the results between the tile and slide levels and the interpretability of the algorithm.
...

References

SHOWING 1-10 OF 45 REFERENCES

Clinical-grade computational pathology using weakly supervised deep learning on whole slide images

A multiple instance learning-based deep learning system that uses only the reported diagnoses as labels for training, thereby avoiding expensive and time-consuming pixel-wise manual annotations, and has the ability to train accurate classification models at unprecedented scale.

Deep Adversarial Training for Multi-Organ Nuclei Segmentation in Histopathology Images

This work uses a conditional generative adversarial network (cGAN) trained with synthetic and real data to train a conditional GAN with spectral normalization and gradient penalty for nuclei segmentation that outperforms conventional approaches, especially in isolating individual and overlapping nuclei.

Automated acquisition of explainable knowledge from unannotated histopathology images

A deep learning based automated acquisition of explainable features from pathology images is reported, and a higher accuracy of their method is shown as compared to pathologist based diagnosis of prostate cancer recurrence.

Artificial intelligence in digital pathology — new tools for diagnosis and precision oncology

A broad framework is provided for incorporating AI and machine learning tools into clinical oncology, with an emphasis on biomarker development, and some of the challenges relating to the use of AI are discussed, including the need for well-curated validation datasets, regulatory approval and fair reimbursement strategies.

Classifying and segmenting microscopy images with deep multiple instance learning

A new neural network architecture is introduced that uses MIL to simultaneously classify and segment microscopy images with populations of cells and it is shown that training end-to-end MIL CNNs outperforms several previous methods on both mammalian and yeast datasets without requiring any segmentation steps.

Development and validation of a deep learning algorithm for improving Gleason scoring of prostate cancer

A deep learning system for Gleason scoring whole-slide images of prostatectomies, developed using 112 million pathologist-annotated image patches from 1226 slides, and evaluated on an independent validation dataset of 331 slides, achieves a significantly higher diagnostic accuracy and trended towards better patient risk stratification in correlations to clinical follow-up data.

Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer

In the setting of a challenge competition, some deep learning algorithms achieved better diagnostic performance than a panel of 11 pathologists participating in a simulation exercise designed to mimic routine pathology workflow; algorithm performance was comparable with an expert pathologist interpreting whole-slide images without time constraints.

Adversarial Stain Transfer for Histopathology Image Analysis

A discriminative image analysis model equipped with a stain normalization component that transfers stains across datasets is designed, which achieves superior results in terms of accuracy and quality of normalized images compared to various baselines.